Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
There are several component technologies ML can teach computers to detect and
within AI all with numerous applications. recognize suspicious behavior and classify
For example, Natural Language Processing alerts following a risk-based approach
(NLP), through which computers process as being of high, medium or lower risk.
human language, is used for translating Applying rules to these alert classifications
payments in payment screening, processing can facilitate hibernation and auto closure
mortgage applications and negative news of alerts, allowing human resources to
screening (see Figure 1). supervise machines that triage these alerts
rather than reviewing alerts manually.
ML is a sub-field of AI where computers This change is illustrated in Figure 2.
are able to learn when exposed to new
data without being explicitly programed. Machines can be taught to recognize,
The potential applications for this within score, triage, enrich, close or hibernate
suspicious activity monitoring and alerts. This can help focus experienced
Transaction Monitoring (TM) are vast. human resources on complex problems
while computers perform simpler tasks
A common TM challenge is the generation for them. By using AI across the entire TM
of a vast number of alerts requiring costly process, alerts can be auto-triaged leveraging
operations teams to triage the alerts. ML; they can be auto-enriched with account,
With increasing customer numbers, and transaction, and customer information from
increasing numbers of transactions, TM external and internal data sources leveraging
alert volumes have increased significantly intelligent automation and NLP; and narratives
at financial institutions (FIs). This increase for Suspicious Activity Reports (SARs) can
in alerts has resulted in an upsurge in the be auto-generated using Natural Language
number of personnel required to triage and Generation (NLG). Institutions leveraging AI in
process them. However, in leveraging ML at this way can reduce dependencies on human
the alert triage stage, alerts can potentially operators to perform routine tasks, reduce
be suppressed, hibernated (set aside for the total time it takes to triage alerts, and
later examination) or even, at some time allow personnel to focus on more valuable
in the future, closed automatically. and complex activities.
DEEP EVIDENCE
Input Hidden Output LEARNING BASED Documents Possible Answers
Layer Layers Layer
RECOMMEN- ML
DATION SYSTEMS Linear Clustering
Data Recommendation ENGINES Regression
Models
PREDICTIVE PRESCRIPTIVE
Available Predictions ANALYTCS AI ANALYTICS Available Assessment
Features Features
Advice
NLP/
NLG
TEXT MINING
Escalated Alerts
SARS
Automation 1%
TXNS
Investigations
Data from
Multiple Automation Auto-enriched Alerts
Sources TM Analysts Closed Alerts
Auto-triaged
Alerts Automation
DQM
and Profile
Refresh
Lack of Lack of a
Straightforward 360-degree
Processes View of the
Customer
CHALLENGES
FOR ML & AI
Limited
Limited
Knowledge of both
Regulatory
Financial Services
Appetite
and ML
Source: Supervisory Guidance on Model Risk Management, Board of Governors of the Federal Reserve System,
the Office of the Comptroller of the Currency
Money Laundering?
+ Hidden
Input Output
Volume CRR
High High
_ Hyperplane
Classification and Artificial Neural
Regression Tree (CART) Support Vector Machines (SVM) Network (ANN)
We have extensive capabilities in this area, We have vendor alliances, both with
including analytics and digital centres of Fintech start-ups2 and with leading firms.
excellence and subject matter specialists to: Accenture consultants meet on a regular
basis with new vendors, help vendors build
Identify and analyze business opportunities their capabilities in the financial crime
Assess the impact of possible solutions prevention space, and evaluate them
(such as updating procedure manuals to deliver greater value to clients. Within
versus adopting ML) the financial crime space, Accenture
has established an effective and efficient
Deliver innovative bespoke and industry
methodology to allow our clients to meet
leading technology solutions, including
regulatory expectations and reduce
implementation of TM systems, AML
volumes of TM and Name Screening
model validation, and operating model
alerts. This methodology is illustrated
transformation to improve the human
in Figure 6 below.
and virtual workforce
Figure 6. Summarized Steps for Rapid Benefits Realization from ML and RPA
Identify pain Analyse available Outline immediate, Enhance and Periodically review
points and required data mid-term and fix data and tune
Identify processes Perform data long-term goals Algorithm and applications for
and procedures quality analysis Identify the "art of vendor selection ongoing suitablility
that can be (e.g. completeness the possible" for Build, deploy User training and
automated ratios) automation and ML and test proofs feedback to
Build a Develop delivery of concept improve usability
consolidated data approach and Model validation and
repository road map regulatory proving
Define underlying Implement and roll
data management out model
framework
Monitor business
benefits